#slow installation on multi GPU RTX5090

5 messages · Page 1 of 1 (latest)

old walrus
#

I created an RTX5090 pod with 7GPU. Ran my installation script. It was terribly slow, after 15min, still not finished. Normally when I used RTX5000 installation is done in a few minutes. Although I did not enable jupyter got a notification that it is taking longer than expected.
killed the runpod, created a new one, same problem. killed runpod again.
created a new runpod RTX5090 with only 1 GPU. This time installation is similar to RTX5000.
Is this expected behavior for a multi GPU machine that installation is taking waaaaay longer? if so, why ?
I would like to use a multi-GPU machine so I can run parrallel multiple model-trainings for my reasearch.

peak cipherBOT
#

To help others find answers, you can mark your question as solved via Right click solution message -> Apps -> ✅ Mark Solution

lilac yachtBOT
dire garnet
#

@old walrus it's possible you might have landed on the same machine over and over again - getting a new pod usually helps you avoid issues with specific pods, but with low supply + multi-GPU this is much more likely. If you can DM me your userID I can credit you

old walrus